Abstract
Background
A key skill for a practising clinician is being able to do research, understand the statistical analyses and interpret results in the medical literature. Basic statistics has become essential within medical education, but when, what and in which format is uncertain.
Methods
To inform curriculum design/development we undertook a quantitative survey of fifth year medical students and followed them up with a series of focus groups to obtain their opinions as to what statistics teaching they want, when and how.
Results
A total of 145 students undertook the survey and five focus groups were held with between 3 and 9 participants each. Previous statistical training varied and students recognised their knowledge was inadequate and keen to see additional training implemented. Students were aware of the importance of statistics to their future careers, but apprehensive about learning. Face-to-face teaching supported by online resources was popular. Focus groups indicated the need for statistical training early in their degree and highlighted their lack of confidence and inconsistencies in support.
Conclusion
The study found that the students see the importance of statistics training in the medical curriculum but that timing and mode of delivery are key. The findings have informed the design of a new course to be implemented in the third undergraduate year. Teaching will be based around published studies aiming to equip students with the basics required with additional resources available through a virtual learning environment.
Introduction
A fundamental component for practising clinicians is reviewing the evidence and making informed choices about suitable treatment for patients. Understanding statistical analyses and accurate interpretation of results is key to efficient clinical practice, and this is recognised by the General Medical Council (GMC). 1 However, these recommendations come with no direction into content and volume of teaching required. As such, statistics teaching varies among UK medical schools.
Despite this recognition, very few previous studies have explored this issue. Miles et al. 2 surveyed practising doctors on their views of undergraduate statistical training and the need for the skills in daily practice. They showed clinicians prefer teaching based around examples of statistical methods used within clinical practice rather than the theory. Postgraduate medical students agreed with the importance of statistical training, but found it difficult, were often unsure of the meaning of statistical terms and struggled with appropriate method choice. 3 Nearly one-third of their subjects indicated a lack of statistics training in their career and suggested more effective training within undergraduate education. A recent study assessing the attitudes of graduate medical students found that even those with numerical backgrounds were still apprehensive and perceived statistics as difficult. 4
At the University of Aberdeen, no formal statistical training exists within the undergraduate medical curriculum and students have had varied statistical experience through student selected modules, electives (small research project) and informal daily drop-in student statistics clinic. Some students undertake a full basic statistics course within the intercalated BSc Medical Science (BMedSci) degree. This course covers basic descriptive statistics, univariate tests (e.g. chi-squared tests, independent t-tests, ANOVA) and an introduction to linear regression.
This mixed methods study aimed to explore the statistical wants and needs of current undergraduates medical students. The results, in combination with previous studies,2–5 will inform the design and delivery of an appropriate statistical course for a large cohort of undergraduate medical students.
Methods
We undertook a concurrent mixed methods study with fifth year medical students at the University of Aberdeen. Given the varied experience, adequate exposure to various clinical situations and potential appreciation of the need for evidence based medicine, fifth year students were deemed ideal to approach to gain opinions about what would be useful, when and in what format.
The study consisted of a quantitative survey and series of focus group discussions (FGD). The latter aimed to obtain an extended understanding of the experiences and perceptions of medical students about statistical teaching received and to identify relevant factors that will contribute to developing statistical teaching. Ethical approval was provided by College Ethics Review Board, University of Aberdeen.
Data collection
The full cohort of 153 fifth year medical students were targeted (academic year 2013/14). The survey was introduced and conducted in a timetabled session of each general practice rotation. With the author’s permission our survey was developed from a previously used survey for practising clinicians.2,6 The questionnaire asked basic demographic information; components students had undertaken in their degree (e.g. clinical contact, numerical analysis) and how relevant they thought medical statistics was within each; when they had been exposed to statistics; in what format and whether it was perceived as relevant. For the elective project, students were asked about their statistical knowledge, whether they struggled and whether they had sought help in planning and analysis. Students were asked about sources of help as well as their opinions on the format and timing of future teaching.
During the survey session, the purpose of the FGD was explained and information leaflets provided and students invited to attend. On the day, written consent was obtained, the process of the FGD was explained and participants were under no obligation to participate and free to leave at any time. A topic guide was used to facilitate the discussions with at least two researchers present at each FGD.
Data analysis
Using IBM SPSS Statistics version 21, 7 all questionnaire variables were appropriately summarised. Appropriate Chi-squared tests (Pearson χ2, continuity correction or Fishers exact test) were used to compare students who completed the BMedSci versus those who had not and second between those who had already completed their elective rotation and those who had not. All FGD were recorded, transcribed verbatim and analysed using ‘Thematic Analysis’. All data were coded for the group and day but otherwise anonymised.
Results – survey
The response rate was 94.7% (145/153). The median (interquartile range) age was 23 (22, 24) years. Just over half were female (n = 79, 55%), with 76% of British nationality (n = 110), 10 (7%) other European and 24 (17%) other nationalities. About one-fifth had completed the BMedSci degree (22%).
Components of practice where statistics needed – N (%).
Statistical training.
Percentages do not add to 100 as respondents were asked to tick all that applied.
Future medical training – helpful modes of teaching.
27% said entirely online would suit them; 56% said no and 17% said maybe.
MyMBChB is the virtual learning environment used by medical students at Aberdeen.
Statistical knowledge was felt to be inadequate for elective projects by 33% with more than 40% saying their knowledge (negatively) influenced their choice of project (online supplementary Table 4). The majority of students (78%) struggled (a little or a lot) with statistical aspects, only 19 (13.4%) sought help at the statistical clinic for planning, but 43.6% with analyses (online supplementary Table 4). Other than the student clinic, the main sources of help reported by students for their elective projects were supervisors (48%), Google (57%), books (26%), YouTube (15%) and, other students (22%).
Receipt of a greater number of previous sessions was associated with the students who had done the BMedSci (p < 0.001) (online supplementary Table 4). Their use of the statistical consultancy service was also significantly higher compared to the non-BMedSci students (56.2% vs. 12.5%, p < 0.001). A greater proportion of the BMedSci students felt their statistical knowledge was adequate for the elective and fewer of them struggled with statistical aspects of the elective. No differences were found in terms of seeking help for planning or analysing their data, and while a greater proportion of the non-BMedSci students said their statistical knowledge influenced their choice of project, the difference was not significant.
Due to the rotation system in fifth year, some of our respondents had not undertaken their elective project, n = 47 (32%). There was no statistical difference in the proportion of the post-elective students who had used the student clinic (26.5%) compared to those who were pre-elective (14.5%), p = 0.271. Having completed their elective, 21.4% felt their statistical knowledge was adequate, but only 10.9% felt it was going to be adequate before embarking on their elective (p = 0.199). A higher proportion of the pre-elective students felt they would struggle a lot with statistical aspects of their elective, 37.0%, compared to 28.9% of those who had done the elective and struggled (p = 0.210). There were no differences between groups for students visiting the clinic when planning the elective (p = 0.625) with 13.4% doing so. Only 9% of those pre-elective thought they would not seek help at the clinic for analysis, but around half (47.8%) thought they would seek a lot. This compares to 79.2% of those having completed the elective not seeking help and only 8.3% seeking a lot of help (p < 0.001).
Results – FGD
In total, five FGDs (36 participants) were conducted with each FGD ranging from three to nine. Four main themes emerged: the need for statistical training in the medical curriculum; lack of confidence; inconsistencies in support and expectations for career. Within each of these themes, several sub-themes were identified as presented in online supplementary Table 5 alongside relevant quotes from students.
Theme 1: need for statistical training in medical curriculum
Most participants indicated a need for basic statistical training within the medical curriculum. They felt/expressed an initial reluctance to learn statistics but that it should be incorporated into the curriculum and made compulsory. There was a discussion about the subject knowledge versus technical and interpretative skills. For example, some were able to undertake a specific calculation, but lacked the theoretical background and skills to interpret them. Some, however, were able to interpret results but had no knowledge to handle and analyse primary data.
Timing of statistical training was an important sub-theme. Students felt the importance of statistics and its application should be emphasised earlier (third year) with the basics statistics taught. This should then be built on and followed up in the fifth year as a refresher course as fourth year was not suitable. This suggestion came with the caveat that very ‘early’ teaching might not be well attended.
Mode of delivery was thought to be important. It was felt that basics should be taught face-to-face, preferably in small groups allowing for interaction and explanation, followed up with on-line material and/or teaching workbooks.
Preparation for elective projects was felt to be inconsistent as whilst all students submit their elective proposals at the same time, they are conducted at different times. This can lead to a long time lag between planning and conducting projects. Students felt that basic statistical knowledge would help in planning project proposals.
Theme 2: lack of confidence
The majority of students expressed a lack of confidence in statistics. Diversity in their backgrounds and lack of previous knowledge/experience contributed to the uncertainty and their need for basic training (theme 1). Students reported letting their uncertainty about statistics influence their elective choice, with some selecting non-numeric projects due to the lack of confidence in handling data.
Theme 3: inconsistencies in support during medical training
The existence of the statistics clinic was not known by many students. While students appreciated this support, some felt that there was not enough time to discuss issues and problems with continuity, as different staff members were available each day. Students felt a huge amount of uncertainty around elective projects and were unprepared for the methodology of their research projects. Students seem to get variable levels of support from elective supervisors, depending on the supervisor’s statistical knowledge and expertise.
Theme 4: expectations for career
Participants realised the need to learn statistics early in their medical training. They felt the importance of learning statistics was not emphasised enough in their earlier days of training, and it was not until experiencing the pressures of electives that this was realised. Most felt that they would need to get involved in research at some point in their career and that statistics would help them with this.
Discussion
The key findings were that there was no obvious consensus on timing of teaching but third year had some advantages. Students preferred small group teaching, were not keen on statistics being examinable, but recognised the need for it to be core to ensure engagement. The format prompted much discussion in the FGD, with students seeing the value of online material as a resource rather than the main mode of delivery. This went against the expectation of the research team, who thought in this digital age, online learning with flexible access would be popular with students.
It was acknowledged that students who had undertaken the BMedSci were at an advantage later in their medical degree and future career. Students felt the importance of statistics to medicine should be stressed much earlier and not on graduating medical school. Very few students sought help in the planning stage of their elective projects a finding echoed by a study in India which found only 31.3% of final year postgraduate medical students sought help in planning. 3
Two medical students from the University of Bristol expressed the opinion that statistical teaching should come as early as possible but with reinforcement thereafter perhaps through the use of an online resource. 8 However they also recognised that if too early in the degree students might not see the relevance, a sentiment echoed in the present study. Students indicated that they knew statistics was important for their careers but that this knowledge was obtained recently and not as an early-year undergraduate. Swift et al. found that student perceptions about what was important for careers were consistent with GMC recommendations,1,6 namely being able to critically appraise published research, perform clinical audit and better understand risk to explain it to patients. In this study, the majority of students recognised the relevance of statistics to these components.
Hannigan et al. recommend that any initial statistical teaching with medical students should focus on the misconception that previous mathematical performance has a relationship with statistical ability. 4 Reducing this anxiety will encourage students to engage with statistics teaching. This anxiety was evident, with students expressing a lack of confidence in their statistical ability and an avoidance of statistics if possible. As a consequence, it is suggested that a statistical glossary is linked to the clinical lectures from the first year, where students will be signposted to statistical terms and additional resources to aid explanation. It is hoped that this light touch approach will allow them to gain confidence with basic statistical concepts used in their everyday practice before any formal statistical teaching is undertaken.
Freeman et al. assessed student performance after two different methods of teaching, one being a standard lecture style theoretical course and another centered around using concepts and examples from the literature utilising a variety of teaching methods and media. 5 The latter proved to be successful, and is corroborated by this study with students indicating that they want some element of face-to-face teaching in smaller groups, followed by a resource (online or paper-based) that is continuously available. So in addition to the provision of statistical terminology early on in the medical curriculum, a short face-to-face course incorporating varying teaching styles and activities is suggested to run at the end of their third year. During this course, terminology used in recent literature will be identified, expanded to show relevance and then developed into basic statistical concepts without too much theory or numerical analyses.
The strength of this study is that it targeted undergraduate medical students while previous studies have focused on practising clinicians. Although based in one institution and statistics teaching differs across institutions, our study findings were in line with other studies, but were of particular pertinence to undergraduate medical training.
Conclusion
This study has provided pertinent views that medical students have about learning statistics, primarily that statistics is important within their medical training, but that timing and mode of delivery are key. As a consequence, a series of elements have been identified as being useful and we are now in the process of developing and implementing an undergraduate statistics course. Some of these may be of interest to other medical schools (UK and elsewhere) and adapted for their particular environment and needs. In the future, we plan to assess the views of students with regard to the success and relevance of these new initiatives, gaining feedback to make improvements.
Footnotes
Declaration of conflicting interests
The author(s) declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: none.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: none.